Implementation Opinions on the Standardized Application and Innovative Development of Intelligent Agents 智能体规范应用与创新发展实施意见

Translated by Claude, edited by Sinocism.com

Agents are intelligent systems with autonomous capabilities for perception, memory, decision-making, interaction, and execution, and constitute an important form of artificial intelligence products and services. With the rapid development of new-generation AI technologies such as large models, agents are accelerating their deep integration with cyberspace and the physical world, profoundly changing modes of production, ways of life, and models of social governance. To implement the State Council's Opinions on Deeply Implementing the "AI+" Action, and to promote the standardized application and innovative development of agents, these implementation opinions are hereby formulated.

智能体是具备自主感知、记忆、决策、交互与执行能力的智能系统，是人工智能产品及服务的重要形态。随着大模型等新一代人工智能技术迅猛发展，智能体正加速与网络空间、物理世界深度融合，深刻改变人类生产生活方式和社会治理模式。为落实国务院《关于深入实施"人工智能+"行动的意见》，促进智能体规范应用与创新发展，制定本实施意见。

I. Basic Principles

一、基本原则

Guided by the imperatives of advancing scientific and technological innovation, enhancing governance capacity, building industrial ecosystems, and improving people's livelihoods, we will adhere to safety and controllability, treating the safety, reliability, and trustworthiness of agents as bottom-line requirements for development, running through the entire process of agent technology R&D, application deployment, and promotion, and effectively guarding against systemic risks. We will adhere to standardized order, adapt to the evolutionary patterns of agent technology, build a governance system that is well-aligned with existing policies and regulations, with industry self-discipline and self-governance, and with clear bottom lines and red lines, and orderly advance the implementation and application of agents. We will adhere to innovation-driven development, strengthen the linkage between theoretical, technological, and engineering innovation, achieve systematic breakthroughs in key agent technologies, improve coordination mechanisms among government, industry, academia, research, and users, build an open and shared agent ecosystem, and enhance industrial innovation vitality. We will adhere to application-driven approaches, focusing on actual needs in scientific research, industrial development, boosting consumption, people's livelihoods, and social governance, leveraging the demonstration effects of typical application scenarios, proceeding from the easy to the difficult and step by step, to promote technology validation, product iteration, and application implementation of agents.

以推动科技创新、提升治理能力、构建产业生态、增进民生福祉为导向，坚持安全可控，将智能体安全、可靠、可信作为发展的底线要求，贯穿智能体技术研发、应用部署与推广的全过程，切实防范系统性风险。坚持规范有序，适应智能体技术演进规律，构建与现有政策法规衔接顺畅、行业自律自治、底线红线清晰的治理体系，有序推进智能体落地应用。坚持创新驱动，加强理论创新、技术创新、工程创新联动，体系化突破智能体关键技术，完善政产学研用协同机制，构建开放共享的智能体生态，提升产业创新活力。坚持应用牵引，重点围绕科学研究、产业发展、提振消费、民生福祉、社会治理等实际需求，发挥典型应用场景示范效应，先易后难、循序渐进，促进智能体技术验证、产品迭代、应用落地。

II. Consolidating the Foundation for Development

二、夯实发展基础

Strengthen the technology foundation, improve the standards system, and lower the thresholds for agent R&D, adaptation, and application, laying the groundwork for enriching agent products and services.

夯实技术底座，健全标准体系，降低智能体研发、适配、应用门槛，为丰富智能体产品及服务奠定基础。

(1) Improving the Technology Foundation

（一）完善技术底座

Strengthen R&D of foundational technologies. Continuously improve the performance of general-purpose foundation models, support the development of specialized models for industry segments, and form a model product matrix adapted to different scenarios and devices. To support agent training and operation, enhance the supply capacity of high-quality datasets. Strengthen technical breakthroughs in agent task understanding, task planning, tool use, long-term memory, mutual recognition and interconnection, and group collaboration, and improve the generalization capabilities of agents.

1.强化基础技术研发。持续提升通用基础模型性能，支持行业发展细分领域专用模型，形成适应不同场景和设备的模型产品矩阵。面向智能体训练与运行，提升高质量数据集供给能力。加强智能体任务理解、任务规划、工具使用、长期记忆、互认互通、群体协同等技术攻关，提升智能体泛化能力。

Improve agent toolchains. Conduct research on underlying agent frameworks, accelerate the development of key components for perception, memory, decision-making, interaction, and execution, and improve toolchains for agent R&D, testing, deployment, and operations. Develop security and governance tools such as adversarial sample detection and behavioral anomaly detection, enhancing capabilities to detect, intervene in, block, and recover from non-compliant agent behavior.

2.完善智能体工具链。开展智能体底层框架研究，加快研发感知、记忆、决策、交互、执行等关键组件，完善智能体研发、测试、部署、运维等工具链。发展对抗样本检测、行为异常检测等安全与治理工具，提升对智能体非合规行为的发现、干预、阻断、恢复能力。

(2) Building Standards and Protocols

（二）构建标准协议

Establish an agent standards system. Formulate guiding documents for agent standardization work, form an agent standards framework, systematically lay out standards systems covering key technologies, important products, data exchange, application scenarios, quality assessment, security assurance, and trustworthy certification, and accelerate the formulation of foundational standards for agent interfaces with software tools, application services, and hardware peripherals. Strengthen the promotion and application of key national and industry standards for agent interconnection such as the Agent Interconnection Protocol (AIP). Support the formulation of mandatory standards in fields such as healthcare, transportation, media, and public security. Encourage enterprises to develop products and services in accordance with relevant standards, improving the standardization of agents. Actively participate in the formulation of international standards.